From:                                         Himel, Thomas M. [thimel@slac.stanford.edu]

Sent:                                           Thursday, September 03, 2009 1:26 PM

To:                                               Walker, Nicholas John; ilc-ops-availability (alias for: Mailmaster, Mailinglist Management)

Subject:                                     Proposed outline for availability task force report in Albequerque and perhaps for the final written report.

 

Below is my proposed outline for the talk:

 

1.       Goal of the task force (find viable solution to have good availability for SB2009 design. Concentrated on single tunnel linac first …

2.       Members/meeting schedule

3.       3 working groups and their assignments

4.       Machine configurations we have studied (central DR still 6 km, long RTML transport lines, e+ source at end of linac, linac in 1 or 2 tunnels, all other areas have accessible support equipment)

5.       Diagram of DRFS scheme

6.       Diagram of klystron cluster scheme

7.       Results are preliminary and WILL change. Mention a few items we are presently unhappy with.

8.       Availsim results of availability as function of energy overhead for 4 configurations: 2 tunnel 10 MW, 1 tunnel 10 MW, 1 tunnel DRFS, 1 tunnel klyClus.

9.       Assumptions. There will be many slides about this and it is the most important part of the talk. It explains what design assumptions we made to get the above results. It will try to explain why we made some of the assumptions and explain that we have not necessarily optimized them yet. A partial list of these assumptions, just to give you the flavor:

a.       DRFS mod anode and DC supplies are redundant so very long MTBF was used

b.      5 month run and 1 month long down (including recovery time) (during which 10% of cryo systems are warmed and have all their bad components repaired.)

c.       Every 2 weeks there are 9 hours of scheduled repairs followed by 15 hours of recovery. There is no limit on number of people used to make repairs. Everything that can be repaired in less than 9 hours gets repaired. This time would also be used for preventive maintenance which is not explicitly modeled but may be needed to attain the very long MTBFs.

d.      Each klystron cluster has 2 spare klystrons and modulators that can immediately be used if 1 fails. Then the klystron or modulator can be replaced while the accelerator is running. This simulated by giving a very long MTBF to the klystrons and modulators.

e.      Big table of MTBFs used including the comparison to other MTBFs that we have found from other labs and the literature. I plan to color code the spreadsheet I showed at our last meeting so it is easy to see how much longer our planned MTBFs are than previous experience.

f.        The LLRF for klyClus is redundant in all parts that handle signals from more than 1 cryomodule. Simulated by very long MTBF.

g.       The long high power waveguides of LLRF do not fail enough to significantly affect the downtime. (i.e. not modeled).

h.      Two people can replace a DRFS klystron in 4 hours including transportation time. The klystrons have an MTBF of xx and nn get replaced on an average down day.

i.         AC breakers and transformers are “lightly loaded” so they get advertised MTBFs and breakers do not trip.

10.   My intention is to list ALL the significant assumptions and run out of time while still showing them. This makes sure we have them documented. I will of course go over the most important ones first.

 

Comments are welcome

 

Tom